CS 471 Operating Systems. Yue Cheng. George Mason University Fall 2017

Transcription

1 CS 471 Operating Systems Yue Cheng George Mason University Fall 2017

2 Outline o Process concept o Process creation o Process states and scheduling o Preemption and context switch o Inter-process communication 2

3 Process Concept o o o Process: a program in execution Process execution must progress in sequential fashion A program is a passive entity, whereas a process is an active entity with a program counter and a set of associated resources Each process has its own address space: Text section (text segment) contains the executable code Data section (data segment) contains the global variables Stack contains temporary data (local variables, return addresses..) A process may contain a heap, which contains memory that is dynamically allocated at run-time ü The program counter and CPU registers are part of the process context 3

4 Process o Introduced to obtain a systematic way of monitoring and controlling program execution o At first: The unit that can be dispatched (scheduled) The unit that owns resources (This view changed later on with the advent of threads ) o A process is an executable program with: associated data (variables, buffers ) execution context 4

5 Processes a) Multiprogramming of four programs b) A conceptual model of 4 independent, sequential processes c) Only one program active at any instant 5

6 OS Requirements for Processes o Utilization OS must interleave the execution of several processes to maximize CPU utilization while providing reasonable response time o Synchronization OS must allocate resources to processes while avoiding deadlock o Communication OS must support inter-process communication and user creation of processes 6

7 Process Creation o Principle events that cause process creation System initialization Execution of a process creation system call by a running process User request to create a process 7

8 Process Creation (cont.) o Parent process creates children processes, which, in turn create other processes, forming a tree (hierarchy) of processes o Issues Will the parent and child execute concurrently? How will the address space of the child be related to that of the parent? Will the parent and child share some resources? 8

9 An Example Process Tree 9

10 How to View Process Tree in Linux? o % ps auxf f is the option to show the process tree o % pstree 10

11 Process Creation in Linux o Each process has a process identifier (pid) o The parent executes fork() system call to spawn a child o The child process has a separate copy of the parent s address space o Both the parent and the child continue execution at the instruction following the fork() system call. The return code for the fork() system call is o zero for the new (child) process o the (nonzero) pid for the parent o Typically, a process can execute a system call like execlp() to load a binary file into memory 11

12 Example Program with fork void main () { int pid; } pid = fork(); if (pid < 0) {error_msg} else if (pid == 0) { /* child process */ execlp( /bin/ls, ls, NULL); } else { /* parent process */ /* parent will wait for the child to complete */ wait(null); exit(0); } 12

13 A Very Simple Shell while (1) { type_prompt(); read_command(cmd); pid = fork(); if (pid < 0) {error_msg} else if (pid == 0) { /* child process */ execute_command(cmd); } else { /* parent process */ wait(null); } } 13

14 Demo: fork 1 What happens to the value of number? 14

15 Results./forkexample1 Running the fork example The initial value of number is 7 PID is 2137 PID is 0 In the child, the number is 49 PID is 0 In the parent, the number is 7 15

16 Demo: fork 2 What happens to the value of number? 16

17 Results./forkexample2 Running the fork example The initial value of number is 7 PID is 2164 PID is 0 In the child, the number is 49 PID is 0 In the child, the number is 49 PID is 0 In the parent, the number is 7 17

18 execl vs. fork 18

19 Results./execlexample Running execl code PID is 2179 PID is 0 In the execl child, PID is 0 Running the fork example The initial value of number is 7 PID is 2180 PID is 0 In the child, the number is 49 PID is 0 In the child, the number is 49 PID is 0 In the parent, the number is 7 In the parent, done waiting 19

20 Process Termination o Process executes last statement and asks the operating system to delete it (exit) Output data from child to parent (via wait or waitpid) Process resources are de-allocated by operating system o Parent may terminate execution of children processes ( e.g. TerminateProcess() in Win32) o Process may also terminate due to errors o Cascading termination when a system does not allow a child process to continue after the parent has terminated 20

21 Reasons for Process Termination o Normal completion o Memory unavailable o Memory bounds violation o Protection error E.g., write to read-only file o Arithmetic error o Time limit exceeded o Time overrun Exception handled by OS Process waited longer than a specified maximum for an event o And a lot more 21

22 Reasons for Process Termination (cont.) o I/O failure o Invalid instruction happens when try to execute data o Privileged instruction o Operating system intervention such as when deadlock occurs o Parent request to terminate one offspring o Parent terminates so child processes terminate 22

23 Process States (simplified) o Running state o Ready state o Blocked state o New state OS has performed the necessary actions to create the process but has not yet admitted the process o Exit state Termination moves the process to this state Tables and other info are temporarily preserved for auxiliary program 23

24 A Five-State Process Model Ready to exit: A parent may terminate a child process 24

25 Process Queues 25

26 Swapping/Suspending o Processes may need to be swapped out to disk This is true with virtual memory! o 2 new states: Blocked Suspend: blocked processes which have been swapped out to disk Ready Suspend: ready processes which have been swapped out to disk 26

27 Additional State Transitions o Blocked --> Blocked Suspend When all processes are blocked, OS will make room to bring a ready process in memory o Blocked Suspend --> Ready Suspend When the event for which it has been waiting occurs o Ready Suspend --> Ready When no more ready process in main memory o Ready--> Ready Suspend When there are no blocked processes and must free memory for adequate performance 27

28 A Seven-State Process Model 28

29 Process Scheduling o The operating system is responsible for managing the scheduling activities A uniprocessor system can have only one running process at a time The main memory cannot always accommodate all processes at runtime for multiprogrammed OS The operating system will need to decide on which process to execute next (CPU scheduling), and which processes will be brought to the main memory (job scheduling) 29

30 Process Scheduling (cont.) o How to pick which process to run? o Scan process table for first runnable? Expensive: Weird priorites (e.g., small pids do better) Divide into runnable and blocked processes o FIFO (FCFS)? Put processes on back of list, pull them from front o Give some processes a better shot at the CPU? 30

31 Process Scheduling Queues o Job queue set of all processes in the system o Ready queue set of all processes residing in main memory, ready and waiting for CPU o Device queues set of processes waiting for an I/O device o Process migration is possible among these queues 31

32 Process Lifecycle 32

33 CPU and I/O Bursts o CPU I/O Burst Cycle Process execution consists of a burst of CPU execution and I/O wait. o I/O-bound process spends more time doing I/O than computations, many short CPU bursts. o CPU-bound process spends more time doing computations; few very long CPU bursts. 33

34 CPU-bound vs. I/O-bound Processes (a) A CPU-bound process (b) An I/O-bound process 34

35 Scheduling Policies o Want to balance multiple goals Fairness don t starve processes Priority reflect relative importance of jobs Deadlines must do X (play video) by certain time Throughput want good overall performance Efficiency minimize overhead of scheduler itself o No universal policy Many variables, multiple objectives, can t optimize them all Conflicting goals (e.g., throughput vs. priority) o We will spend a whole lecture on this topic 35

36 Preemption o Can preempt a process when kernel gets control o Running process can vector control to kernel System call, page fault, illegal instruction, etc. May put current process to sleep e.g., read from disk May make other process runnable e.g., fork, write to pipe o Periodic timer interrupt If running processes used up quantum, schedule another o Device interrupt Disk request completed, or packet arrived on network Previously waiting process becomes runnable Schedule if higher priority than currently running process o Changing running process is called a context switch 36

37 Example of Preemption Act I/O request OS preempts P2 OS preempts P1 Disk interrupts Process 2 s execution Each square: one CPU cycle 1: Process 1 2: Process 2 37

38 Context Switch 38

39 Context Switch (cont.) o Very machine dependent. Typical things include Save program counter and integer registers (always) Save floating point or other special registers Save condition codes Change virtual address translation o Non-negligible runtime cost! Save/restore floating point registers expensive Optimization: only save if process uses floating points May require flushing TLB (memory address translation hardware) Usually cause more cache misses (switch working set) 39

40 Process Communication o Mechanism for processes to communicate and to synchronize their actions. o Two models Communication through a shared memory region Communication through message passing 40

41 Communication Models Message Passing Shared Memory ü Previously, in a distributed system, message-passing was the only possible communication model. However, remote direct memory access (RDMA) technique bridges this gap by providing remote memory access through network. 41

42 Communication through Message Passing o Message system processes communicate with each other without resorting to shared variables o A message-passing facility must provide at least two operations: send(message, recipient) receive(message, recipient) o With indirect communication, the messages are sent to and received from mailboxes (or, ports) send (A, message) /* A is a mailbox */ receive (A, message) 42

43 Communication through Message Passing o Message passing can be either blocking (synchronous) or non-blocking (asynchronous) Blocking Send: The sending process is blocked until the message is received by the receiving process or by the mailbox Non-blocking Send: The sending process resumes the operation as soon as the message is received by the kernel Blocking Receive: The receiver blocks until the message is available Non-blocking Receive: Receive operation does not block; it either returns a valid message or a default value (null) to indicate a non-existing message 43

44 Communication through Shared Memory o The memory region to be shared must be explicitly defined o Using system calls in Linux: shmget creates a shared memory block shmat maps an existing shared memory block into a process s address space shmdt removes ( unmaps ) a shared memory block from the process s address space shmctl is a general-purpose function allowing various operations on the shared block (receive information about the block, set the permissions, lock in memory, ) o Problems with simultaneous access to the shared variables o Compilers for concurrent programming languages can provide direct support when declaring variables (e.g. shared int buffer ) 44

45 Shared Memory Example

46 Shared Memory Example 46

47 sharechild.c Src 47

48 Output./share.exe shared memory segment attached at address 0x10b74c000 In the execl child, PID is 0 memory is 0x10b74c000 From Child==>Hi there CS 471! * In the parent, done waiting In Parent==>Hi there CS 471! * 48